If you are looking for the areas of a new algorithms missing, I think Ted recently published a list of things sought after.
I for myself would very much like to see SVM things done at scale. Another feature request from me is hierarchical ckustering, if you like a new challenge. The challenge for any new algorithm is to figure a parallelization technique that is usually not a part of a previously publicized matter. So if you prefer a ground up work, be prepared to do some of that heavy lifting first and have it reviewed. What we found out(ok, let me speak for myself, i found out) is that the best strategies are being created in collaboration and review. Having new unique angles at the problem from different people creates most balanced solution, although the actual coding jira task is usually driven by a single contributor, subject to a subsequent review. I usually try to get some assurance I am heading in the right direction first. E.g. 817 would benefit greatly if somebody could double check my math (in particular, median propagation under BB' part) and perhaps even simulate it in matlab (something that you seem to be very skillful at). You already helped me a lot with your simulation of m Omega. Thanks. On Nov 28, 2011 2:26 PM, "Raphael Cendrillon" <cendrillon1...@gmail.com> wrote: > Hi Ted, > > I think the difficulty I have is in identifying areas to contribute that > the community will find useful. > > If I understand correctly at this stage the major algorithms are in place > and the focus is on polishing the existing code rather than adding large > amounts of new functionality. > > With this in mind it seems the best thing to do is find an existing module > to work on. So I'm wondering if there are any existing module maintainers > that wouldn't mind taking someone new under their wing? > > The cost of this is that it requires a bit more coordination, but in the > long term I think it will lead to better quality code and a more successful > project for everyone. > > On Nov 28, 2011, at 2:02 PM, Ted Dunning <ted.dunn...@gmail.com> wrote: > > > What are you finding difficult? > > > > Us? > > > > The process? > > > > The concept? > > > > What can we do to make this easier? > > > > On Mon, Nov 28, 2011 at 1:57 PM, Raphael Cendrillon < > > cendrillon1...@gmail.com> wrote: > > > >> I would like to get involved in contributing to the code, although I'm > >> finding this quite difficult. > >> >
